GreenPill S.10 Ep.9: ElizaOS, FOSSRPG, and the Future of Open-Source AI Agents with Shaw
10 snips
Feb 7, 2026 A wide-ranging chat about building ElizaOS and shipping agent tooling through a bear market. They unpack agent-native games, OTC trading experiments, and ports to Python and Rust. The conversation dives into open-source mindshare, community dynamics, and how agents reshape work, ownership, and funding models. They end on cultural shifts needed for meaningful, coordinated building.
AI Snips
Chapters
Transcript
Episode notes
Shipping Fast, Then Rewriting
- Shaw shipped Eliza V2 then rebuilt core tooling because multi-agent workflows and consumer tooling created new problems.
- He ported Eliza to Python and Rust and used it in games, trading desks, and embedded experiments.
Validate Agents With Games First
- Use games and simulated environments as cheap benchmarks to probe what agents can do beyond base models.
- Build lower‑hanging experiments (trading game, prediction game, RuneScape sim) to surface real agent capabilities.
Hype PRs Don't Make Sustained Communities
- The initial Eliza contributor surge was driven partly by token/hype incentives that produced low‑quality collaboration.
- Shaw now favors a smaller core of committed builders and commissioned contributors.